The Safety Work Nobody Sees: Why Frontline Reality Rarely Matches the Procedure

Every organisation has a version of how work is supposed to happen. It exists in procedures, risk assessments, method statements, permit systems, training materials, operating manuals, and management expectations. The task is planned, the hazards are identified, the controls are established, and competent people are expected to follow the process.

Then the work starts.

Equipment is not exactly where it was expected to be. A delivery arrives late. The weather changes. A contractor sends a different crew. Access is restricted. A tool is unavailable. Another activity creates an unexpected interface. Information is incomplete.

A supervisor is dealing with several competing priorities, and the team still needs to complete the work safely.

People adapt.

In most organisations, thousands of these adaptations occur every day. They rarely result in incidents. In fact, many of them are the reasons work is completed successfully despite the variability, uncertainty, and imperfections of real operations.

Yet much of traditional safety management pays remarkably little attention to this work.

We spend enormous effort documenting how work should happen, but considerably less effort understanding how work actually happens.

That gap matters because some of the most valuable information about operational risk exists inside it.

Work as Imagined and Work as Done

The distinction between “work-as-imagined” and “work-as-done” has become increasingly important within human factors, resilience engineering, and contemporary safety thinking.

Work-as-imagined describes how an organisation expects work to be performed. It is the version represented by procedures, plans, organisational charts, risk assessments, schedules, and management-system requirements.

Work-as-done is what actually occurs when people interact with real equipment, real environments, real constraints, and real operational pressures.

The two are rarely identical.

That does not automatically mean something has gone wrong. No procedure can anticipate every combination of circumstances that employees may encounter. Even highly standardised operations contain variability, and people continually make small adjustments to keep work moving.

A procedure may assume that a particular valve is easily accessible, for example, while an operator knows that reaching it requires working around temporary equipment installed during maintenance. A lifting plan may assume an exclusion zone can be established exactly as shown on a drawing, while the lifting supervisor discovers another contractor has occupied part of the area. A project schedule may assume one sequence of work, while delayed materials make another sequence more practical.

The organisation sees the procedure.

The workforce experiences the reality.

Understanding the difference between those two perspectives is essential if we genuinely want to understand risk.

Procedures Cannot Describe Every Situation

Procedures are necessary. In high-risk industries, they are essential.

The problem begins when organisations assume that documenting a process means the operational problem has been solved.

A procedure is essentially a prediction. It describes how the organisation believes a task should be completed under an anticipated set of circumstances. Good procedures incorporate experience, engineering knowledge, regulatory requirements, lessons learned, and established controls. But they are still written before the next job begins.

The people performing the work operate in the present.

They encounter the actual equipment condition, the actual weather, the actual workload, the actual staffing level, and the actual conflicts between simultaneous activities. They are required to reconcile what was planned with what exists.

Sometimes the adjustment is trivial. Sometimes it introduces risk.

The important question is whether the organisation can see the difference.

If our safety processes only confirm whether people followed the procedure, we may completely miss the operational conditions that made following the procedure difficult in the first place.

Adaptation Is Not Automatically Non-Compliance

Traditional safety thinking can be uncomfortable with adaptation because deviation from a prescribed process is often interpreted as non-compliance.

Sometimes that interpretation is correct. Deliberately bypassing a critical control or ignoring an essential safety requirement cannot simply be dismissed as a normal operational adaptation.

But not every difference between procedure and practice represents reckless behaviour.

People routinely compensate for weaknesses elsewhere in the system. They clarify incomplete instructions. They coordinate conflicting activities. They rearrange work when equipment becomes unavailable. They identify hazards that were not captured during planning. They slow activities down when conditions deteriorate. They ask experienced colleagues for advice when a situation falls outside the procedure.

These behaviours are rarely recorded as safety interventions, yet they contribute enormously to successful performance.

If organisations treat every adaptation as a behavioural failure, employees quickly learn an important lesson: do not tell management how the work really gets done.

That creates a significant problem.

The organisation may believe it has achieved procedural compliance when it has actually achieved procedural silence.

The Hidden Safety System

Most experienced workplaces develop an informal layer of operational knowledge that never appears in the formal HSE management system.

People know which equipment is temperamental. They know which areas become difficult to access during certain activities. They know which procedures are technically correct but cumbersome in practice. They know which contractors require closer supervision. They know where communication routinely breaks down and which jobs almost always require additional coordination.

Much of this knowledge is passed from person to person through conversations, mentoring, toolbox meetings and experience.

In many cases, this informal knowledge is helping the organisation manage risk.

The problem is that it may remain invisible to senior management.

This becomes particularly important when experienced personnel leave. If the organisation has relied on individuals to compensate for weaknesses in its formal systems, their departure can remove controls that management did not even realise existed.

A procedure remains in the document management system, so the organisation assumes nothing has changed.

Operationally, however, significant capability may have disappeared.

This is one reason organisations should pay much greater attention to the practical knowledge of experienced employees and contractors. Their value is not simply that they know the rules. Often, they know where the rules are insufficient.

Why Traditional Audits Can Miss the Gap

Audits are an essential part of effective HSE management, but traditional compliance audits can struggle to reveal work-as-done.

An auditor asks whether a procedure exists. It does.

The auditor checks whether employees have been trained. They have.

The auditor samples completed documentation. The records are available.

The organisation receives a satisfactory result.

But none of those checks necessarily establishes whether the procedure accurately reflects how work is being performed.

A more revealing audit would spend time observing the activity itself. It would ask workers where the job becomes difficult, which steps require judgement, where conditions regularly differ from the plan and what they do when that happens.

This does not replace compliance assurance. It strengthens it.

Compliance tells us whether expected controls are present. Operational learning helps us determine whether those controls make sense in the conditions where they are being applied.

We need both perspectives.

Stop Asking Only “Why Didn't They Follow the Procedure?”

The difference becomes particularly important after an incident.

One of the first questions frequently asked during an investigation is, “Why wasn't the procedure followed?”

That may be a legitimate question, but it can also prematurely frame the investigation around individual compliance.

A better starting point is to understand the conditions surrounding the work.

What made sense to the people involved at the time? What information did they have? What pressures were present? What equipment was available? What assumptions were being made? Was the procedure practical under those conditions? Had the task been performed this way previously without consequence?

These questions do not remove accountability.

They improve the quality of it.

There is an important difference between understanding why an action occurred and excusing the action. Organisations need to be capable of doing the first without automatically doing the second.

If an investigation concludes that someone “failed to follow the procedure” without understanding why, it may identify the immediate deviation while leaving the underlying operational conditions completely untouched.

The corrective action then becomes predictable: retrain the worker, reissue the procedure, conduct a toolbox talk, and remind everyone to comply.

The organisation has technically responded to the incident while learning very little from it.

The Danger of Successful Deviations

Perhaps the most difficult adaptations to identify are those that work.

Imagine a team discovers a quicker way of performing a task. It deviates slightly from the approved method but creates no obvious problem. The job is completed successfully.

The same approach is used again.

Again, nothing happens.

After enough repetition, the alternative method can become a normal practice. New employees learn it from experienced workers. Supervisors may become aware of it.

Eventually, the way the task is actually performed can differ significantly from the way the organisation believes it is being performed.

Success reinforces the behaviour.

This is one pathway through which the normalisation of deviation can develop. The absence of negative consequences is interpreted as evidence that the practice is safe.

But the absence of an incident does not necessarily mean the risk is adequately controlled.

Sometimes the organisation is succeeding because its controls are effective.

Sometimes it succeeds because people are adapting intelligently. And sometimes it is succeeding because the conditions required for failure have simply not aligned yet.

A mature safety organisation needs to distinguish between the three.

Leaders Need to Get Closer to the Work

Understanding work-as-done cannot be achieved entirely from dashboards, monthly reports or executive meetings.

Leaders need exposure to the realities of operations.

This does not mean another ceremonial safety walk where senior leaders arrive wearing new PPE, ask workers whether they are working safely and leave twenty minutes later.

The objective should be learning rather than inspection.

A useful leadership conversation sounds different. What makes this job difficult? What regularly changes from the plan? Which part of the procedure is hardest to apply? What equipment or information would make the work safer? Where do you rely on experience rather than written guidance? If you could change one thing about this process, what would it be?

Those questions communicate something important: management is interested in understanding the system, not simply catching people doing something wrong.

The quality of the answers will depend heavily on trust.

If workers believe admitting that a procedure is difficult to follow will lead to disciplinary action, they will tell leaders what they think leaders want to hear. If they believe raising operational problems leads to improvement, they are far more likely to explain what is really happening.

This is where psychological safety becomes operationally relevant. It is not about creating an environment where nobody is challenged. It is about creating enough trust for inconvenient information to travel upwards.

Bring the Procedure and the Work Back Together

Recognising the difference between work-as-imagined and work-as-done does not mean abandoning procedures or allowing everyone to choose their own way of working.

Quite the opposite.

The objective should be to continually bring the formal system and operational reality closer together.

That requires organisations to observe work, engage employees, capture operational learning, and update processes when the evidence shows they no longer reflect reality.

If workers consistently need to adapt a particular step, investigate why.

If a procedure regularly requires informal clarification, improve it.

If supervisors have developed effective local practices, determine whether those practices should be incorporated into the formal system.

If the prescribed method genuinely represents the safest approach but people are unable to follow it, identify the conditions preventing compliance rather than simply repeating the requirement.

Sometimes the answer will be better training. Sometimes it will be better equipment, improved planning, additional resources, clearer responsibilities, or redesigned work.

And sometimes the procedure itself needs to change.

The important point is that the organisation learns.

Safety Happens in the Gap

The most interesting part of a management system is often not what the procedure says.

It is what happens when reality does not match the procedure.

That is where people make decisions. It is where experience becomes important. It is where competing objectives are reconciled, where weak systems are compensated for, and where risk can either increase or be successfully controlled.

Traditional HSE approaches have spent enormous effort trying to eliminate variability from work. In many situations, standardisation remains exactly the right objective.

Critical controls, isolations, lifting requirements, confined-space entry and other high-risk activities need clearly defined boundaries that should not be casually adapted.

But organisations also need to recognise that variability is an unavoidable feature of complex work.

The challenge is therefore not simply to demand that reality conform perfectly to the procedure.

The challenge is to understand where reality differs, why it differs and whether those differences are helping or undermining the control of risk.

The safest organisations are not necessarily those where nobody ever deviates from what was imagined.

They are organisations that are good at seeing how work is really being done, learning from the difference and continually improving the system around the people doing the work.

Because some of the most important safety work happening in an organisation will never appear on a checklist, dashboard or audit report.

It happens quietly, every day, as people adapt to reality and keep the operation working.

The question for leaders is whether the organisation is learning from it.

References and Further Reading

Erik Hollnagel. Safety-I and Safety-II: The Past and Future of Safety Management. Ashgate, 2014. Introduces the distinction between traditional approaches focused primarily on preventing adverse outcomes and approaches that also seek to understand how everyday work succeeds.

Erik Hollnagel. FRAM: The Functional Resonance Analysis Method – Modelling Complex Socio-Technical Systems. Ashgate, 2012. Explores performance variability and the interactions between functions within complex systems.

Sidney Dekker. The Field Guide to Understanding 'Human Error'. CRC Press. Examines human performance, local rationality and the importance of understanding why actions made sense to people within the circumstances they faced.

Health and Safety Executive (HSE). Human Factors: Managing Human Failures. Guidance addressing organisational, job and individual factors that influence human performance and the limitations of approaches focused primarily on individual error.

Energy Institute. Human and Organisational Factors resources. Industry guidance and resources addressing human performance, organisational conditions and the integration of human and organisational factors into safety management.